Dear All
At today's EGEE operations meeting more information was given about
changes to the site report timing:
- There was a proposal to always report for a "civil" week (mon-sun).
This would lead to:
- a week window from Monday morning till late Sunday
- sites can edit each day (except when ROCs can), and can finish their
part on Monday & Tuesday morning
- ROCs have from Tuesday noon to Wednesday noon to report
- Wednesday afternoon and Thursday morning can be used to extract the
ROC issues and prepare for the meeting.
I was slightly concerned that this will mean only site remarks on tests
up until 3-4 days before the meeting will be raised but looking at the
way things currently work this may not be a real issue. Important last
minute (i.e. early week) problems can still be added under site comments
- the bit that really counts for the meeting.
In parallel with this there has also been a move to full use of the OPS
VO for monitoring. Due to the changeover there has been a period where
the default SAM page showed dteam results (often all green) and not ops
results which often had problems. The CIC reports also used the dteam
SFT tests creating more confusion. This mixed status is a one off and
should explain the background to comments such as this from one UKI
site:
"The move to OPS is welcome, but the default tests shown on the SAM
website need to be updated to reflect this. It's only now that I realise
we've been failing some significant tests for quite some time.
As this is the first time we have been presented with these results I am
marking failures as non-relevant for this week. Looking at the old SFTs
we had a near perfect week. But now the OPS/SAM tests have flagged us as
failing all week. Clearly this is nonsense as we've been full all week
processing CMS and LHCb jobs!"
Tying up the status according to SAM vs the real state of a site
according to job load is something we will be investigating over the
coming months.
Please let me know (raise a trouble ticket if you want a definite follow
up) if you notice any continued oddities in the SFT results vs the CIC
portal reports. If you do want to call attention to something please
send as many details as possible so we can properly investigate.
Further topics discussed at the ops meeting can be seen on the agenda
page: http://agenda.cern.ch/fullAgenda.php?ida=a063464. Minutes should
be available shortly.
regards,
Jeremy
|